Alex Bovee

Although a bit overused these days, the adage “Sunlight is the best disinfectant” seems so fitting for Castlight Health that I just can’t resist. The three year old Health 2.0 startup based in San Francisco is taking a page directly from Justice Brandeis’ playbook by shining some light on the historically opaque world of healthcare services pricing and quality. Their web based platform, only available through employers for now, allows users to search for doctor’s visits, lab work, and various other services, with corresponding out-of-pocket cost calculation and quality metrics. With a handful of companies already signed up for the service, a killer management team, and a great technology, they’re going to be a company to watch in the next few years.

Read the rest of my review here on TechCocktail

Log management may not be the sexiest problem in IT today, but it sure is an important one – and Loggly wants to solve it. The company is building a cloud based logging service that takes the pain out of log collection, management, and processing and offers it on a subscription basis. The product is currently in private beta, but will soon be available to the general public (btw –love the signup page).

Managing and parsing log data is a problem that has existed since the early days of computers – though now, on a bigger scale than ever before. Today, even a basic application with moderate usage can generate many megabytes of data on a daily basis on aggregate across the web server, OS, and application layers. This problem only been exacerbated recently as companies have focused more and more on collecting as much data as possible on user interaction and traffic.

Read the rest of my review here on TechCocktail

I just finished listening to Brian Chesky’s talk for Startup School from a few months ago. If you haven’t watched it yet, you’d be well advised to do so (find it here). Aside from being hilarious and entertaining, it also provided a unique view into the will, determination, and scrappiness that it takes to start a company. There were lots of great lessons, but one thing in particular caught my attention: AirBnb’s initial user acquisition model was not scalable.

An Intro to AirBnb

AirBnb is an online marketplace that helps users list and rent their most valuable personal asset: the space in their house (well, now treehouses & boats as well). The company has grown tremendously over the last few years, thanks in part to a great user experience and solid value proposition, but also due to some great marketing. They also have an incredible iPhone app (from what I hear at least, I’m on Android).

As a two-sided network, AirBnB needs two distinctly different users to go to their site: a lister and a renter. They can’t create a transaction without those two parties.

The Unique Challenge of Two Sided Networks

Having been in a handful of investor presentations, a question that inevitably comes up with marketplace businesses is: how will you grow the network? It’s a valid question, but for a startup with just an idea or barely functional prototype, it can also be difficult to answer. The founders of AirBnb had a difficult time building their network, and it’s no surprise: marketplace businesses are notoriously difficult to build. Here are just a few reasons:

• They require coordination: If you build up one side of the network, and not the other, you don’t make money (imagine eBay with lots of buyers and no listings).
• They can be difficult to figure out: it’s hard to get one user acquisition model right, let alone two. Often times, buyers and sellers for marketplaces must be reached through different marketing channels. This means optimizing two sets of user acquisition models. There has been lots of great thinking on how to seed one side of the market first.
• They can be costly: Marketing to both sides of the marketplace costs money. Iterating through your acquisition models costs money. Two-sided networks cost money to build. 

Essentially, building a two-sided network or marketplace business to critical mass means nailing product-market fit, marketing message, and marketing execution for two different groups of users!

The Importance of Non-Scalable User Acquisition

The AirBnb story caught my attention not because it’s the first time I’ve heard of marketplace companies using non-scalable user acquisition methods, but because it’s probably the fifth. Another great example that comes to mind is SitterCity.com. SitterCity is a Chicago based company that has built a two-sided network that helps connect parents to babysitters. Genevieve Theirs, the CEO, tells a story of the early days of Sittercity when initial user acquisition came from papering dorms around her campus with fliers for the service. Was that scalable or cost effective? Probably not. And although I’m sure she had no intention of moving forward with this model, it helped her get in touch with her customers and get immediate feedback on the service as well as provided a solid seed for the network.

Similarly, Brian Chesky and the AirBnb team, after working restlessly to build their two-sided network, were coached to get out and just start meeting with customers. And then something miraculous happened. Through this highly un-scalable process, they signed up new users. But, even more importantly, they learned about their customers and their unique set of needs. A few business model tweaks later combined with the seeded network that they had built, and the company left ramen profitable in the dust.

Get Out and Build your Network

Long story short, if you’re looking to build a two-sided network business, don’t be afraid to just get out, roll up your sleeves, and make something happen. As Steve Blank says in his Customer Development process: “there are no facts in the building; get out and meet your customers”.

I’ve read a lot of great posts recently on the future of wireframes and if the low-fidelity mockup should go the way of the dodo.  In one post, the writer advocates using the technology framework itself for creating a mockup – or in other words, producing web-ready HTML/Javascript/CSS (and corresponding graphics from Photoshop or Fireworks) to create a semi-close-to-real-world model for the end site.  Another post has the perspective that low fidelity wire-framing tools do not provide the modeling capabilities for the interactivity and dynamic nature of web 2.0 sites.  A central theme to both posts is that iteration takes time and that the low/medium fidelity wireframing process adds an unnecessary layer to the design process. 

In general, I agree on both fronts.  I can certainly understand how each of these arguments is valid given a set of circumstances.  However, I do think there are cases in which it still makes sense to produce low-fidelity wireframes.  I’ve outlined these below, but first, some definition:

The “Throw-Away” Mockup – Defined

I’m defining the throw away mockup as a wireframe which, for all intents and purposes, is completely unusable to produce anything that will go in production.  This includes mockups that are just rough drafts done on paper, perhaps designed in powerpoint, or in a rapid low or medium fidelity wireframing tool such as Balsamiq.  The outputs from these tools cannot be transformed directly into HTML, but rather are good for socialization, brainstorming, and rapid iteration.  Their final drafts must later be translated into a production format (images, CSS, & HTML) which can then be staged to a developer.

When would you use Low-Fidelity Design Tools in the Design Process?

There are several reasons why you might want to use low-fidelity design tools to rapidly prototype a design.  Here are a few:

• When the Design Concept is Still a Rough Idea: If you don’t have a good idea of how something should work, it’s probably not smart to dive right into Photoshop. At the early stages of an idea, you’re going to go through lots of iterations.  Redesigning user experience and optimizing interactions takes time, thought, and lots of rework.  Unless you’re a supreme whiz at photoshop and can burn through iterations quickly, you’re probably better off starting on paper and then migrating the design to something more industrial strength later.
• When the site is extremely sophisticated: I recently drafted up some Balsamiq designs for a Health 2.0 consumer web company.  If you’ve ever touched the healthcare space, you’ll know it’s very complicated.  There are lots of data elements and translating user flows into simple, intuitive interfaces requires thought and iteration.  Creating production ready mockups in this case would almost certainly be a waste of time.  Throwing out a fully drafted HTML page with CSS and Photoshop template to boot is costly (both to your time and sanity).  Using a low-fidelity mockup tool helps keep the cost of this rework down to a minimum.  Designing Interactive has a good, short write up on this same point.
• When feedback is critical, but comes after design: In a real-time collaborative environment, feedback is immediate.  Drafting up screen elements and flows on a whiteboard lends itself to instant iteration.  However, there are many times when design is done independently and feedback comes after the fact.  In this environment, collaborators may be much less likely to tear down your design if they believe you’ve put a lot of effort into creating it.  Paper, Powerpoint, and Balsamiq mockups look ugly… and that’s the point.  No one will be afraid to tell you your design stinks if it looks hand drawn.
• When product owners are not proficient with high fidelity design tools:  In most startups, the product is owned by the founders.  Sometimes founders are great with design tools, but that’s just not always the case.  Furthermore, becoming proficient with sophisticated graphics tools is not an easy task and probably isn’t a learning curve you want to tackle when you’re already biting off more than you can chew.  In this case, it’s better to work with low-fi mockup tools to create good enough mockups that you can then collaborate on with designers.

You may notice that three of these four reasons could also be put succinctly: “when you believe you’ll have lots of iterations” (the 4th reason is skill centric).  Before jumping in to create high fidelity wireframes, use this litmus test to think about how much iteration it might take to get the design right and select the right tool accordingly.

New Jersey based startup Opinionaided wants to help you get advice and opinions in real-time. Their platform allows you to pose questions to the world or to your personal network, and crowdsource “thumbs up” or “thumbs down” feedback instantly. Want to know whether to purchase the iPad with 32GB or the cheaper, younger brother? Or, whether it’s worth heading to the movies to watch the new Harry Potter movie? Snap a picture, type in a question, and submit the data using Opinionaided’s iPhone app or website. Within minutes, feedback starts rolling in from friends, family, and the world…

Read the rest of my review here on TechCocktail

Hotpot, Google’s new product for local search and recommendations, was released a few days ago to mixed reviews.  Some were great (Mashable) and others were less than stellar (Wired, GigaOM, SEW).  Personally, I like the idea behind Hotpot and think there’s potential.  Personalized recommendations for places to go to, businesses to patronize, and interesting things to do is something that would add a lot of value to my life.  Furthermore, since Google has a plethora of personal and user behavior data, they’re uniquely positioned to provide accurate and interesting recommendations.  Gmail, for example, is a holy grail of information about my personal preferences, hobbies, and passions.

Given the mixed reception, I sat down and thought up a few ideas on how Google could improve Hotpot.  Here they are:

Idea 1 – Call out the Benefit

aka: What’s the Catchphrase?

Hotpot needs to clarify the value proposition and benefit for consumers.  I felt like this was missing both at signup time and within the application.  The signup page says: “Start rating now with Hotpot”.  That sounds like work.  Yuck.  What’s in it for me?  Once you’re inside the application, it doesn’t get much better.  Hotpot’s catchphrase is … well, there isn’t one.  Under the search bar it says: “by Google Places”.  The first flip card says: “Recommendation for here or to go”… I’m not sure what that means.  Conversely, a glance at Yelp’s homepage yields the following catchphrase: “Real People, Real Reviews”.  That doesn’t sound like work and I understand the value that’s in it for me.  Hotpot needs to do a better job of explaining what it is and how it’s going to make my life better.

Idea 2 – Provide Immediate Value:

“Now eventually you might have dinosaurs [recommendations] on this dinosaur tour [recommendation engine]?”  Dr. Malcolm, Jurassic Park

Google has access to a plethora of my information via gmail, buzz, and a host of many other services that I use.  But once I was on the Hotpot site, I sat and scratched my head wondering exactly where the recommendations were on this “recommendation” engine.  It wasn’t intuitive even after rating a handful of businesses.  I spent my first 5 minutes clicking star ratings for several of the businesses that I recognized, but nothing seemed to get better. No additional businesses popped up.  No changes in my existing card set happened.  After a handful of reviews, I wondered what the heck I was doing. It took me logging back in a few days later before I actually saw any recommendations.  Hotpot should start providing users with valuable recommendations from the first click.

Idea 3 – Focus on User Experience:

“Think in Flows, Not Screens” Jason Putorti

Right now, the only flows in Hotpot consist of signing into the application, clicking reviews, and clicking businesses.  This experience feels awkward and stuffy - like I’m interacting with a databot whose job is to suck valuable review information out of me and shove it into a large database.  Hotpot is Google places with a weak social (friend connect), game mechanic (leaderboard), and input/recommendation engine (flipcards) layered on top.  The application could be dramatically improved by focusing on user engagement and flow.  Hotpot could initially ask a user to select a category of business they would like to explore.  The user could then scroll through that category and spend time on things that are relevant and meaningful to them.  That would also help avoid situations where a user is presented with completely irrelevant places.  After six reviews, I was surprised to get a card for: “Encinal High School”:

I’m not in high school.  I just graduated with my MBA from Chicago.  My little one is 2 years old and unless she turns out to be a Doogie Howser, MD, she won’t be in high school anytime soon either.  Hotpot needs to design flows that maximize value for the user and engage them relevantly.

Excited to See What’s Next

Despite some shortcomings, Hotpot did a lot of things right on this release.  Signing up for the service was painless and easy.  It’s layered on top of Google places and uses Google authentication, so if you already have a Google account, it’s just a few extra clicks to get started.  I was also really impressed with the review aggregation functionality.  They’ve sourced reviews from several different locations and boiled all of that information into an easy to understand, 5-star based review system.  Awesome.

All that said, Hotpot is certainly no Yelp killer and will need some work to better engage users.  Some of these changes will be small tweaks, such as clarifying their marketing message for the application.  Others will take substantial investments in time and money.  It will be interesting to see what Hotpot does next.

If haven’t used TypeKit yet, you should definitely try it out.  TypeKit allows you to add creative and professional looking type faces to your website - beyond the basic fonts supported by most browsers/OSs.  The service packages the fonts and sends them directly to the end user’s browser, enabling your website to provide a consistent visual experience across browsers and operating systems.  It’s provided on a freemium basis with a step up in cost based on website traffic and number of sites supported.  The lowest level service allows for up to two fonts on one unique domain with a maximum 25,000 unique page views of traffic per month.  For the average blogger, this is probably all you need.  To get started with an account, just visit TypeKit and use your domain to signup.

There are a couple of ways to use TypeKit on Tumblr.  If you’re lucky, your theme will support TypeKit out of the box.  I use a WooTheme for my Tumblr blog.  It has a lot of UI supported configurability out of the box including the ability to type in my TypeKit ID under the “Appearance” tab:

To get the TypeKit ID for your website, launch the “Kit Editor” on TypeKit.com for your website and then click “Embed Code”.  Your TypeKit ID is listed in the bottom right hand corner of the popup:

If your Tumblr theme does not support adding a TypeKit ID through configuration, you’ll have to add the HTML code directly.  To do so, copy the included HTML in the above “Embed Code” section.  Then, go to customize your blog on Tumblr, select the “Theme” dropdown, switch to “Use Custom HTML”, and paste the copied HTML into the <head> section of the document:

If you load your website and it looks good - you’re golden.  If not, you’ve got some troubleshooting to do.  CSS is incredibly tricky to troubleshoot, especially when multiple stylesheets are being loaded from different domains.  When troubleshooting Tumblr/TypeKit integration, your problem is probably going to fall into one of these two buckets:

• Domain Match Errors
• Styled Elements Errors

1. Domain Match Errors

These occur when the configured domain in your TypeKit editor is not correct.  For me, the problem occurred because I was using TypeKit on a sub-domain on my alexbovee.com domain.  If you have a subdomain such as “blog.yoururl.com” setup, you’ll need to be sure that TypeKit references this domain correctly.  You can check this by going to the Kit Editor and then clicking the “Kit Settings” link:

Here, you’ll see the domain you setup before.  If your blog is on your root level domain, just make sure the domain is “yoururl.com”.  If you are using a sub-domain for your blog mapped to your tumblr domain, you’ll need to be sure your domain is either “blog.yoururl.com” (where blog => your sub-domain) or “*.yoururl.com”.  The wild-card domain selector will enable using this TypeKit on your root and sub-domains.

Note: Be sure to select “Publish” to save these changes and have them pushed out to TypeKit content servers.

2. Styled Elements

If your issue isn’t a domain configuration issue, it’s probably a CSS application issue.  Checking that elements are styled correctly is a little more difficult.  My go-to for this is to use the Firefox add-in Firebug to inspect the elements.  If you don’t have Firebug, it’s an absolute go-to for troubleshooting all things browser, HTTP, and CSS related by allowing you to visually and textually inspect elements on a page, sniff HTTP packets, and more.  To check an element to make sure it’s styled correctly with your font, you can right click it on the page, then select “Inspect Element”:

This will open Firebug with that HTML element selected.  Next, click the “Computed” box in the Style viewer on the right-hand side:

This will give you the computed CSS style for the selected element.  If you don’t see your font (as named in your TypeKit) in the “font-family” section, something is going wrong with your element styling.  From here, check the following:

• Make sure the font-family name you use in your styling corresponds to the style name in the typekit.  You can find this style name by going into the TypeKit Editor and clicking the “Advanced” link next to the style in the “Selectors” section.

• You can use custom “Selectors” in the TypeKit editor to apply TypeKit fonts using CSS that’s sent to your browser by TypeKit.  You add these selectors to HTML tags or classes in the TypeKit editor under the “Selectors” section.  For example, enter “h1” into the selector box and click “Add” to apply this font to all “h1” elements on your site.  Be sure to click “Publish” to save these changes.
• When a TypeKit is loaded, it loads a CSS file from the TypeKit content servers.  You can see this file by clicking the “CSS” tab in Firebug, and then selecting it from the drop down (it will be the only css file from the domain “typekit.com”).  This is a great way to determine if your fonts are being correctly loaded from the TypeKit server.  If you just added custom selectors, and they’re not showing up, chances are that the file has not yet been pushed out or you didn’t “Publish” it.  On the other hand, if no elements are in this CSS file, there’s a good chance that your domain is configured wrong in the TypeKit editor.

Good luck!

Just got back from a “lunch with the Skyara founders” event that I discovered and purchased through their website, Skyara.com.  Well worth the $20.

If you haven’t checked it out yet, Skyara is a local-activities website where you can find exciting things to do around the city.  For the moment, that includes activities as varied as marksmanship training with an army ranger, Barrista lessons, and a most-likely-to-be depressing date with this guy.  Being a bit of an experience junkie myself, I love this idea.  I’m always looking for fun things to do and cool places to check out in San Francisco.  I find the Bay Area to have a high explorability factor as the neighborhoods are so different, the area culturally rich, and the geography diverse.  So, I guess that SF isn’t just a great place to start a company; it’s also a great place to start a company that sells things to do.

The product is easy to use and impressive considering these guys have only really been riffing on it for a couple of months.  You just go to the website, find something that looks interesting, pay for it, and show up on time.  Pretty simple really.  As the product gets built out further and there are more activities, I’d also love to see a recommendation engine.  You can image with information pulled in from your social graph, Skyara could be incredibly effective at helping you part ways with your money.  As for the team, they all quit their corporate jobs after getting into I/O ventures, and moved out here to start the company.  True entrepreneurs, they’re cash efficient like you wouldn’t believe and are working out of the I/O office in the Summit Cafe on Valencia.

What most intrigues me about Skyara is that it could be a great platform to help niche business operators or people with creative hobbies or services to find customers - a localized Etsy for experiences instead of crafts.  Right now the website has a lot of traditional outdoorsey offerings such as paintballing, cycling trips, and hiking expeditions.  Although those activities are certainly valuable, they’re probably not going to make the site that interesting.  More important are the unique, one-off activities - a 30 minute punch the clown session, lunch with the AirBnb engineers (to discus time travel), etc.  Those will be the key to keeping users engaged and coming back to the site for more.  I’m sure there will be some organic growth for these listings (as the service attracts more eyeballs, more people will think up things to list) but growing the volume substantially will require intervention.  They may want to take a page out of Trazzler’s playbook by using contract writers and contests to produce content.

With a good product in place, I’m looking forward to the sure-to-be creative marketing these guys will do.  They received great press last week in TechCrunch, but with so much noise in this startup environment, you could always use more.  I’m thinking a contest to win a really unique experience combined with a splash of social/game mechanics could really drive registrations.  I.e. the experience is won by lottery, but for every friend you refer who signs up to the site, you get an additional entry.  You could spend a week brainstorming cheap guerrilla marketing activities, but I’d probably start by coming up with a catchy headline.  How about: “Skyara: Do Something Awesome”.

Follow Skyara on Twitter here

Yellowstone is the most amazing place I’ve ever been - both in terms of wildlife and landscape.  Jon and I started out heading directly to Old Faithful to meet Alexis and Emily (friends from B-School).  Not even half an our into the park from the Eastern entrance, we came to a dead stop behind a line of stopped cars.  As we were in a bit of rush to make our rendezvous, we pulled around to pass them.  Bam.  Grizzly - walking along the side of the road.  This was a highlight for the drive, but not the only cool sight.  We also passed about 2 or 3 bison that were near or in the middle of the road.



Once at Old Faithful, we met with with Emily and Alexis, and proceeded to grab a seat by the geyser.  Apparently it blows every half an hour or something like that (plus or minus 5 minutes maybe) - hence the name Old Faithful.

From Old Faithful, we decided to drive up the west side of Yellowstone to see some of the thermal features and then loop back around to Old Faithful for the night.  The north western side of Yellowstone is known as the thermal feature area and the entire landscape is full of them.



If you’re looking to plan a trip, definitely get a Yellowstone map or guide with the natural features listed on it so you can plan your trip and check out as much as possible.  We were really lucky in being able to get a hotel room at Old Faithful for the night because we were persistent and called frequently.  However, we were literally only able to get the room for one night and it wasn’t exactly the Ritz Carlton.  I’ve heard of people calling ahead a year in advance to book a place to stay.

We knew Day 2 was going to be just as crazy as Day 1 in terms of how far we needed to drive.  So, we decided to head out early and put as many miles past us as possible in the hopes of making it to Yellowstone before end of day.  We had briefly looked at a map and noticed the Badlands national park was on our way.  We passed it around noon and decided to stop and check it out.

The Badlands actually ended up being quite amazing.  I’m not exactly sure how a rock formation like this is created, but it reminded me of the underwater caverns in Mexico called cenotes.  The rocks looked bizarrely carved up as if water had taken away a mold that previously rested around it.  I wish we could have spent more time there, but we were pressed for time and needed to make it to Rushmore quickly so we could get to Wyoming by nightfall.  So, our stop consisted of little more than looking at a few of the awesome formations followed by some token touristy pictures.

We continued on I-80, and after a short time, landed at the exit for Mt. Rushmore.  To be perfectly honest, I was not entirely sure of what to expect from our stop there.  I had heard from several people that it was very disappointing and “smaller” than what’s to be expected.  I was actually pleasantly surprised.  The drive to the mountain took us through a small town and eventually around a bend where we came about Mt Rushmore head-on.  Parking was easy, which immediately put me in a great mood.  The walk to the monument was not necessarily life altering but it certainly wasn’t disappointing.  The whole area is grandly presented with lots of flags and a nice walk to base the mountain.

We left Mt. Rushmore after about 30 minutes (I’m not sure you need much more time than that).  From there we headed west through Thunder Basin National Grassland and then through Bighorn National Forest.  I had talked to a few people and consequently heard that the Bighorn National Forest was absolutely amazing.  It did in fact blow me away.  The entire drive was full of great canyons, amazing mountains, and beautiful trees and landscapes.  If we had the time, I would have loved to stay, explore, and spend a few nights there camping.

Ultimately we were able to make it to Cody - the hometown of Buffalo Bill.  Apparently there is also a rodeo every night there during the summer (which we missed).